Author Topic: A New AMS Thread  (Read 52159 times)  

Offline Cassie Leigh

  • Status: Scheherazade
  • *****
  • Posts: 1684
    • View Profile
Re: A New AMS Thread
« Reply #700 on: April 13, 2018, 11:52:31 AM »
But in the example you gave with Keywords A & B, why would you have to choose one over the other? (Both keywords are getting impressions and clicks and sales, so Amazon isn't going to look at either as bringing down the ad.)
This is all based on my personal theory of how these things work, which could be wrong...

This is how I think of it. Say I have a budget of $5 a day for an ad. Every time Amazon displays my ad it has to reserve an estimated amount I might pay for a click against my budget. For example, if I might pay 50 cents on average for a click (just an easy number to work with) then that means within my budget I can have ten clicks per day. So Amazon isn't going to show that ad to a million people at once. It could show it to 10,000 people at once, perhaps, but it's not an unlimited number of people who will see that ad at any given time because of my budget constraints. There has to be some sort of optimization formula at play behind the scenes that works with average click rates and bids and budgets to make sure an ad doesn't drastically overspend its budget at any given time.

Which to my thinking means there are only so many slots for displaying that ad at any given time. And I'd far rather that Amazon displayed my ad on a keyword that takes 1/10 the number of impressions to get a sale, because I'll get that sale faster and maybe some organic visibility that will drive more sales. Plus, I won't have the slower keyword using up one of my potential slots that could be going to the faster keyword.

Obviously, there's some balancing you do. I don't just go with the absolute best keywords. But I do try to use my bidding to drive Amazon to focus on my  higher-performing keywords and I will also shut down those keywords that are taking too much "attention" and not delivering sales.


8 Pen Names. Genres: Non-fiction, Speculative Fiction, Romance.
Blog | Video Courses

Offline Cassie Leigh

  • Status: Scheherazade
  • *****
  • Posts: 1684
    • View Profile
Re: A New AMS Thread
« Reply #701 on: April 13, 2018, 12:07:32 PM »
I would think Amazon does not take clicks or not into account. They just place the ad - clicks or not.

If you read through the two main AMS threads on here you'll see a lot of discussion about ads that stop delivering. They get a lot of impressions but not enough clicks or sales and eventually they just stop running altogether. And I do know for a fact that AMS will shut down an ad that's getting too many impressions without clicks, because they did it to one of my early ads. (It was a 1 click to 18,000 impressions ratio on that one.)

Also, if I have keywords that generate sales for me, why would I want to keep a keyword running that doesn't? It doesn't make sense for me to do that. Lots of impressions with few clicks and no sales means poor targeting in my experience. I'm not reaching my potential audience if no one is even willing to click through and see what my book is about. (I had this with my fantasy series when I targeted Harry Potter. Lots of impressions, very few clicks, no sales. I learned it was much better for me to target Sarah J. Maas.)

Plus too, I have zero reviews which spells rank-new-book and/or author. These get passed over quickly I think.  People look for what appears, at-a-glance, what they could know would be good. More reviews = more interest.

I'm not so sure reviews are that important. I have a non-fiction title that's done very well in AMS. When I first started running that title it had no reviews and it didn't get its first review for three months and yet AMS drove consistent sales to that book for that entire period. And you'd think with non-fiction the quality might be an even bigger factor because you want to make sure the person knows what they're talking about. And maybe the book sold better after getting a review or two, but AMS still drove sales even without the reviews. (Especially with SP ads where the ad shows in search results, I think it's less of an issue.)

Obviously, the stronger the book overall with respect to cover, ad copy, title, blurb, reviews, and competitive pricing the better it'll do in general and with AMS. But my goal with running AMS is to generate sales. So if I'm not getting sales from my ads then I know I need to fix one of the items in that list or I need to better target my ad.


8 Pen Names. Genres: Non-fiction, Speculative Fiction, Romance.
Blog | Video Courses

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #702 on: April 13, 2018, 02:58:56 PM »
If you read through the two main AMS threads on here you'll see a lot of discussion about ads that stop delivering. They get a lot of impressions but not enough clicks or sales and eventually they just stop running altogether. And I do know for a fact that AMS will shut down an ad that's getting too many impressions without clicks, because they did it to one of my early ads. (It was a 1 click to 18,000 impressions ratio on that one.)

Also, if I have keywords that generate sales for me, why would I want to keep a keyword running that doesn't? It doesn't make sense for me to do that. Lots of impressions with few clicks and no sales means poor targeting in my experience. I'm not reaching my potential audience if no one is even willing to click through and see what my book is about. (I had this with my fantasy series when I targeted Harry Potter. Lots of impressions, very few clicks, no sales. I learned it was much better for me to target Sarah J. Maas.)

I'm not so sure reviews are that important. I have a non-fiction title that's done very well in AMS. When I first started running that title it had no reviews and it didn't get its first review for three months and yet AMS drove consistent sales to that book for that entire period. And you'd think with non-fiction the quality might be an even bigger factor because you want to make sure the person knows what they're talking about. And maybe the book sold better after getting a review or two, but AMS still drove sales even without the reviews. (Especially with SP ads where the ad shows in search results, I think it's less of an issue.)

Obviously, the stronger the book overall with respect to cover, ad copy, title, blurb, reviews, and competitive pricing the better it'll do in general and with AMS. But my goal with running AMS is to generate sales. So if I'm not getting sales from my ads then I know I need to fix one of the items in that list or I need to better target my ad.

Thanks Cassie. I am not near a 3 month on ads for the book - only a month out really into the ad.  It is fiction and a short read. But you can't tell that from the ad - you have to click. I really think reviews help but only as a visual element for a buyer or KU. 

There is a strange mix of things that impact the buyer or download. 

As for keywords shutting down on you --- if you say so.  Seems Amazon is not holding up its end of the deal if your bid is high and you win the bid. So AMS is nixing your bid for a more likely sell over your ad it seems - if that is what they are doing.

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #703 on: April 13, 2018, 04:23:13 PM »
This is all based on my personal theory of how these things work, which could be wrong...

This is how I think of it. Say I have a budget of $5 a day for an ad. Every time Amazon displays my ad it has to reserve an estimated amount I might pay for a click against my budget. For example, if I might pay 50 cents on average for a click (just an easy number to work with) then that means within my budget I can have ten clicks per day. So Amazon isn't going to show that ad to a million people at once. It could show it to 10,000 people at once, perhaps, but it's not an unlimited number of people who will see that ad at any given time because of my budget constraints. There has to be some sort of optimization formula at play behind the scenes that works with average click rates and bids and budgets to make sure an ad doesn't drastically overspend its budget at any given time.

Which to my thinking means there are only so many slots for displaying that ad at any given time. And I'd far rather that Amazon displayed my ad on a keyword that takes 1/10 the number of impressions to get a sale, because I'll get that sale faster and maybe some organic visibility that will drive more sales. Plus, I won't have the slower keyword using up one of my potential slots that could be going to the faster keyword.

Obviously, there's some balancing you do. I don't just go with the absolute best keywords. But I do try to use my bidding to drive Amazon to focus on my  higher-performing keywords and I will also shut down those keywords that are taking too much "attention" and not delivering sales.

Interesting take. You are targeting and trying to out-guess an algo. 

But I would think the time frame of the algo, as it spins its life, would be seconds at the largest, while your attempts could take days to figure out an attack. An algo would determine if an ad failed and spin the next as quick as it could place it. True, there is a limit to how many it would determine to place not to bust the budget.

So if this is true, bump the budget and lower the bid and watch the effect? Do impressions increase?

edit : so you are trying to find that sweet spot where AMS is hunting placement for what you want. Give AMS a chance to spend your budget by finding the right people.  Make it worth their while maybe?  Does an algo have a worthwhile-level it hunts for?  :D

You seem to sort of suggest that.  Thanks.
« Last Edit: April 13, 2018, 04:28:54 PM by Max 007 »

Offline BigSlimJim

  • Status: Dr. Seuss
  • *
  • Posts: 17
    • View Profile
Re: A New AMS Thread
« Reply #704 on: April 14, 2018, 09:31:27 AM »
Seems Amazon is not holding up its end of the deal if your bid is high and you win the bid. So AMS is nixing your bid for a more likely sell over your ad it seems - if that is what they are doing.

From what I've determined are the best minds here, Amazon's main 'deal' is not with the self-pubbing AMS advertiser. It's with their browsing customer. If you're getting impressions but not interesting the buyers as evidenced by clicks and, ultimately, sales, you're wasting Amazon's shoppers' time, for which they will rightfully put an ad to sleep. An AMS rep was quoted way back on one of these threads confirming that your success is determined by a combination of bid amount AND click through rate. That's the real deal. They don't just sell advertising real estate to the highest bidder if their shoppers aren't interested in and buying that bidder's products over time. And he doesn't NEED to be the highest bidder if the algo thinks his keywords are generally more relevant to sales than higher rollers'.

Granted, the robot's decisions can seem dysfunctional and enormously capricious (even more so if you're an Advantage Central member with access to matched search terms on all keywords with clicks). Keywords that you think are sure-fire winners will be subject to insane matches to search terms. But since we can't argue with an unthinking machine, and we know it reacts badly to words that perform badly once given their dubious chance-- to the point of punishing the ad in general, what is there to do but kill and/or replace our under-performing keyword-darlings?

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #705 on: April 14, 2018, 01:54:01 PM »
This is all based on my personal theory of how these things work, which could be wrong...

This is how I think of it. Say I have a budget of $5 a day for an ad. Every time Amazon displays my ad it has to reserve an estimated amount I might pay for a click against my budget. For example, if I might pay 50 cents on average for a click (just an easy number to work with) then that means within my budget I can have ten clicks per day. So Amazon isn't going to show that ad to a million people at once. It could show it to 10,000 people at once, perhaps, but it's not an unlimited number of people who will see that ad at any given time because of my budget constraints. There has to be some sort of optimization formula at play behind the scenes that works with average click rates and bids and budgets to make sure an ad doesn't drastically overspend its budget at any given time.

Which to my thinking means there are only so many slots for displaying that ad at any given time. And I'd far rather that Amazon displayed my ad on a keyword that takes 1/10 the number of impressions to get a sale, because I'll get that sale faster and maybe some organic visibility that will drive more sales. Plus, I won't have the slower keyword using up one of my potential slots that could be going to the faster keyword.

Obviously, there's some balancing you do. I don't just go with the absolute best keywords. But I do try to use my bidding to drive Amazon to focus on my  higher-performing keywords and I will also shut down those keywords that are taking too much "attention" and not delivering sales.


Thanks Cassie. I can see the logic to that.

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #706 on: April 14, 2018, 02:03:13 PM »
On one AMS book with a $1/day budget I was recently charged $2.69/day. On another with a $2/day budget I was charged $3.38/day. I wrote AMS and asked what was up. Here was the reply:

Quote
Your average daily budget is the amount youre willing to spend per day over a calendar month. For example, if you set your average daily budget to $100, you may receive up to $3,100 worth of clicks in that calendar month (assuming a full 31-day month).

Daily budgets can be exceeded as a result of a rapid accumulation of clicks on a campaign as the campaign approaches the daily limit. During high traffic seasons, it's recommended that you adjust your daily budgets and cost-per-clicks (CPC) to accommodate increased traffic and avoid overages. If overages to a daily budget occur, you will be charged only the maximum budget of the day. The difference will not be charged.

I track with everything except:

Quote
If overages to a daily budget occur, you will be charged only the maximum budget of the day. The difference will not be charged.

Which contradicts what they said in the first paragraph.

Anybody know what's up with that?


"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline Cassie Leigh

  • Status: Scheherazade
  • *****
  • Posts: 1684
    • View Profile
Re: A New AMS Thread
« Reply #707 on: April 14, 2018, 02:08:09 PM »
In their guidance on the site they do say that your daily budget is an average for the month which is in line with that first response. But when they bill you you'll see that there's a section of the bill that includes credits for overdelivery of your ad. I assume there's some threshhold they use. So if your budget is $1 and one day it's $1.10 they're not going to refund that. But if they miscalculate on your ad and have a $2.50 day, then they'd refund the difference on that one. My last bill I had two ads that spent about $80 each that had refunds of approximately $5 credited back to them but another that spent $180 that only had 20 cents credited back to it.


8 Pen Names. Genres: Non-fiction, Speculative Fiction, Romance.
Blog | Video Courses

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #708 on: April 14, 2018, 03:53:08 PM »
In their guidance on the site they do say that your daily budget is an average for the month which is in line with that first response. But when they bill you you'll see that there's a section of the bill that includes credits for overdelivery of your ad. I assume there's some threshhold they use. So if your budget is $1 and one day it's $1.10 they're not going to refund that. But if they miscalculate on your ad and have a $2.50 day, then they'd refund the difference on that one. My last bill I had two ads that spent about $80 each that had refunds of approximately $5 credited back to them but another that spent $180 that only had 20 cents credited back to it.

I tried an experiment : I raised the budget to $6/day yesterday from $1/day and left the bid at $0.25.  I now have 3x the impressions I had before accumulated over two days. So in one day I had tripled the impressions with the increased budget. And I have a click for $0.25.  No buy. 

With KU I guess you won't know till it is read (let me know if this is true). I had 60 impressions and now 180 (gain of 120 in one day over the 60 from previous two days and one click). "Time Travel" is the keyword that got it.  The ad copy says, "... FREE to Kindle Unlimited!" at the end too.

But anyway, I am right (or you are Cassie). AMS wants to spend your money as best a bot can. And it does seem budget constraints come into play. Too many keywords and a low budget chokes the bot. Give it a chance and it will hunt better for you I think.

2nd experiment : I shutdown all keywords but the one I had a hit on.  It only had 20 impressions and 1 click.  But it seems to be a high bid word - "Time Travel".  The other 160 impressions were just impressions.  From the past, the keywords were extremely high on impressions and no clicks (8000 impressions). So turn them off. They confuse the bot I think.

The premise here is concentrate the bid and budget to one word that I know works.  This may give the bot more ability to get more impressions and more of them to the front of the carousel.  The reader has no idea the bot is directing their attention and hope my book appears first or near the front on the carousel. So I may have to go to a bid of 35 cents.

The problem is ... the book is only 78 pages.  So I need conversion or my ROI is really going to suck.  I can add in my second book to the first and make it 200 pages and keep the price at 99 cents.  So I will get more money from KU than I would a sell. I write for KU, it is a hobby ...  ;)

But I bet it never reaches the budget - it is a tool to focus the bot. As a hobby I don't mind spending a few dollars a month if this can get traction.

If this works, I schedule a 5 day $0 promo, leave AMS like it is and hope for traction just using AMS as the promo tool and raise the bid to 35 cents too. 

If that works, I ram in the second book as fast as I can in a month into the first.  2nd edition and notify previous readers if I can through Amazon.

The idea : spend a little money for exposure and traction.

Anyone see any flaws?  I would be interested to know your thoughts.

Offline ParkerAvrile

  • Status: Jane Austen
  • ***
  • Posts: 359
    • View Profile
    • Parker Avrile's Steamy Gay ROmance
Re: A New AMS Thread
« Reply #709 on: April 14, 2018, 06:36:50 PM »
On one AMS book with a $1/day budget I was recently charged $2.69/day. On another with a $2/day budget I was charged $3.38/day. I wrote AMS and asked what was up. Here was the reply:

I track with everything except:

Which contradicts what they said in the first paragraph.

Anybody know what's up with that?

I am also being charged triple what I authorized the last few days. 
Visit my website for free reads & giveaways for lovers of steamy m/m romance https://parkeravrile.wordpress.com/blog/

Offline LilyBLily

  • Status: Arthur C Clarke
  • *****
  • Posts: 2054
    • View Profile
Re: A New AMS Thread
« Reply #710 on: April 14, 2018, 08:09:20 PM »
I am also being charged triple what I authorized the last few days.

That's why these ads can be scary. We have to assume that what we authorize for one day will be taken by AMS as license to spend it every day of the month. We're just lucky AMS doesn't actively try to spend an entire month's budget all in one day, the way it has been known to with PD ads.

 

Offline Philip Gibson

  • Status: Arthur C Clarke
  • *****
  • Posts: 2473
  • Gender: Male
  • Laos (Moved from Yorkshire)
  • Philip and Cookie the Dog
    • View Profile
    • Hashtag Histories
Re: A New AMS Thread
« Reply #711 on: April 15, 2018, 06:31:52 AM »
Granted, the robot's decisions can seem dysfunctional and enormously capricious (even more so if you're an Advantage Central member with access to matched search terms on all keywords with clicks). Keywords that you think are sure-fire winners will be subject to insane matches to search terms. But since we can't argue with an unthinking machine, and we know it reacts badly to words that perform badly once given their dubious chance-- to the point of punishing the ad in general, what is there to do but kill and/or replace our under-performing keyword-darlings?

"Punishing the ad in general" implies that you think AMS evaluates the overall performance of a campaign - not just the individual performances of its component keywords.

I don't see how this would work.  Is it not the individual keywords + ad copy + bid + history that AMS evaluates?

What if there had been social media during important historical events?
Philip Gibson | Hashtag Histories | Word By Word Graded Readers for Children

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #712 on: April 15, 2018, 10:50:48 AM »
In their guidance on the site they do say that your daily budget is an average for the month which is in line with that first response. But when they bill you you'll see that there's a section of the bill that includes credits for overdelivery of your ad. I assume there's some threshhold they use. So if your budget is $1 and one day it's $1.10 they're not going to refund that. But if they miscalculate on your ad and have a $2.50 day, then they'd refund the difference on that one. My last bill I had two ads that spent about $80 each that had refunds of approximately $5 credited back to them but another that spent $180 that only had 20 cents credited back to it.

Thanks Cassie. I don't know though. Like I said in the previous post I was charged $2.69 for a $1 dollar limit and I've had a lot of those types of over-charges and I've been doing this for over a year and I've never received a refund of any amount.

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #713 on: April 15, 2018, 10:54:28 AM »
Is increasing a bid a Catch-22? You maybe get a sale on a keyword whose ACPC is only .08. Great, since it's costing me so little I up my bid. Then the ACPC jumps right up too. If I keep increasing the bid, the ACPC is suddenly .38. My spend is rapidly accumulating and the sales aren't keeping pace. My great opportunity is now a money drain.

What's a sane way to increase bids?

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline BigSlimJim

  • Status: Dr. Seuss
  • *
  • Posts: 17
    • View Profile
Re: A New AMS Thread
« Reply #714 on: April 15, 2018, 12:38:19 PM »
"Punishing the ad in general" implies that you think AMS evaluates the overall performance of a campaign - not just the individual performances of its component keywords.

I don't see how this would work.  Is it not the individual keywords + ad copy + bid + history that AMS evaluates?
AMS shutting down impressions on campaigns based on overall performance is not something I've experienced myself yet, because I just recently got my feet wet and then paused my ads which had low CTRs before this might have happened. I was going from what I thought I'd often heard here, most recently Cassie's Reply #701:

If you read through the two main AMS threads on here you'll see a lot of discussion about ads that stop delivering. They get a lot of impressions but not enough clicks or sales and eventually they just stop running altogether. And I do know for a fact that AMS will shut down an ad that's getting too many impressions without clicks, because they did it to one of my early ads. (It was a 1 click to 18,000 impressions ratio on that one.)

A couple of questions on your four-part equation, individual keywords + ad copy + bid + history: Does Amazon actually judge our ad copy other than to initially approve or reject it? And when you say history, are you referring to past CTR and sales?


Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #715 on: April 15, 2018, 02:31:33 PM »
Is increasing a bid a Catch-22? You maybe get a sale on a keyword whose ACPC is only .08. Great, since it's costing me so little I up my bid. Then the ACPC jumps right up too. If I keep increasing the bid, the ACPC is suddenly .38. My spend is rapidly accumulating and the sales aren't keeping pace. My great opportunity is now a money drain.

What's a sane way to increase bids?

There's not a sane way.  Especially if the book is smallish. A KU borrow for me gains me 35 cents on pages read. A click costs me 25 cents.  I will run into a negative ROI in two clicks with no buy or borrow. Book is 99 cents - so a buy is 35 cents to me.  But the keyword that gets clicks seems to have a really high bid of 25 cents a click just to be there.

The only solution is to increase the book size and price, which I can do easily - I have the second smallish book almost done - result is the book will be 3 times its size now. Really it is one book anyway; it will be about 250 pages.  Get some ROI wiggle room and set the price at $2.99.

Plus to make matters worse, if you want to be on the front of the carousel, the bid has to be higher I bet to outbid others for that spot. 

So, the best I can tell ... this is all a hobby just to get read.  For most people at KBoards that seems to me to be the case.  If your spend is $1 a day but you allow for $6 (to entice the AMS bot), you spend $30 a month to have a few people read your book (if that many). 

That is a hobby folks, spend a little money to have some fun, if that gives you a thrill and some pleasure to do just that. Big deal on ROI if you can live with the cost.

That may be the only way to keep it sane in the beginning.

Down the road if you can rank into the top 100 the tide may turn.  Plus if you have some follow on books too, this may help some.

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #716 on: April 15, 2018, 03:09:10 PM »
AMS shutting down impressions on campaigns based on overall performance is not something I've experienced myself yet, because I just recently got my feet wet and then paused my ads which had low CTRs before this might have happened. I was going from what I thought I'd often heard here, most recently Cassie's Reply #701:

A couple of questions on your four-part equation, individual keywords + ad copy + bid + history: Does Amazon actually judge our ad copy other than to initially approve or reject it? And when you say history, are you referring to past CTR and sales?

Yeah. Me too. 

It would be odd to me to think the AMS bot has accumulative data slots to tally and collect to determine that an ad should be shut down for each and every ad the turns on and off and changes for each keyword.  That's a lot of data to collect and spin off of behind the scenes. 

It could be there is some herd effect going on. After awhile the keywords that worked in the past naturally become less effective - readers see the same books appearing and they migrate to a different path if they don't pick any books.

AMS really acts like a "Carnival Barker" trying to get someone's attention.  But the keywords should start working again after some time - days - weeks - whatever.

Offline Philip Gibson

  • Status: Arthur C Clarke
  • *****
  • Posts: 2473
  • Gender: Male
  • Laos (Moved from Yorkshire)
  • Philip and Cookie the Dog
    • View Profile
    • Hashtag Histories
Re: A New AMS Thread
« Reply #717 on: April 15, 2018, 07:29:57 PM »
AMS shutting down impressions on campaigns based on overall performance is not something I've experienced myself yet, because I just recently got my feet wet and then paused my ads which had low CTRs before this might have happened. I was going from what I thought I'd often heard here, most recently Cassie's Reply #701:

A couple of questions on your four-part equation, individual keywords + ad copy + bid + history: Does Amazon actually judge our ad copy other than to initially approve or reject it? And when you say history, are you referring to past CTR and sales?

1.  If it is the case that AMS evaluates a campaign as a whole, then it would make sense for us to remove all non-performing keywords, leaving only the high performers.  That would greatly improve the performance of the ad as a whole and produce better results.  I just don't see it working that way.  I think it's down to the individual keywords and their associated components (ad copy + bid + history) being evaluated irrespective of the performance of the other keywords in the ad they happen to share and the overall performance of the ad in which they sit.

I would like to be wrong about this.

2.  I believe ad copy is where 'relevance' comes in.  It is certainly judged for relevance when being approved, but I would think its relevance would continue to be evaluated as time progresses.  But evaluated per associated keyword, not evaluated per ad overall.

3.  Yes, 'history' being past CTR and sales.

That said, I'm no less in the dark than most here.  My assumptions may all be wrong.  However, somebody, somewhere, does know exactly how the AMS algorithms work.

But, so far, they're not talking.

Philip
« Last Edit: April 15, 2018, 08:11:25 PM by Philip Gibson »

What if there had been social media during important historical events?
Philip Gibson | Hashtag Histories | Word By Word Graded Readers for Children

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #718 on: April 16, 2018, 10:05:16 AM »
There's not a sane way.  Especially if the book is smallish. A KU borrow for me gains me 35 cents on pages read. A click costs me 25 cents.  I will run into a negative ROI in two clicks with no buy or borrow. Book is 99 cents - so a buy is 35 cents to me.  But the keyword that gets clicks seems to have a really high bid of 25 cents a click just to be there.

The only solution is to increase the book size and price, which I can do easily - I have the second smallish book almost done - result is the book will be 3 times its size now. Really it is one book anyway; it will be about 250 pages.  Get some ROI wiggle room and set the price at $2.99.

Plus to make matters worse, if you want to be on the front of the carousel, the bid has to be higher I bet to outbid others for that spot. 

So, the best I can tell ... this is all a hobby just to get read.  For most people at KBoards that seems to me to be the case.  If your spend is $1 a day but you allow for $6 (to entice the AMS bot), you spend $30 a month to have a few people read your book (if that many). 

That is a hobby folks, spend a little money to have some fun, if that gives you a thrill and some pleasure to do just that. Big deal on ROI if you can live with the cost.

That may be the only way to keep it sane in the beginning.

Down the road if you can rank into the top 100 the tide may turn.  Plus if you have some follow on books too, this may help some.

Thanks Max. Yeah, definitely a hobby for me at this point. Although some people invest big-time in it. And yeah, the 99 cents thing is just about impossible to even break even.

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #719 on: April 16, 2018, 10:19:06 AM »
Trying to figure this out.

4/15/18


4/16/18

So 4/15/18 and before I was getting an average cost of .14/click. So I raised my bid and then my very next click cost me .72? Wow. Kind of surprised. Sure, I moved up in the carousel but (with all due respect to the author) this author was hardly Stephen King.

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline Gregg Bell

  • Status: Arthur C Clarke
  • *****
  • Posts: 2050
  • Gender: Male
  • Itasca, Illinois
    • View Profile
    • greggbell.net
Re: A New AMS Thread
« Reply #720 on: April 16, 2018, 10:22:34 AM »
So it seems to me, if you're content to be on the back pages you can have a maintain a modest bid, but if you increase your bid enough to move up to the front page, you're likely to get hit with a big cost for a click. (Even with smallish authors.) Would this be accurate?

And yet, sometimes, even with a high bid, the ACPC stays the same (with more clicks).
« Last Edit: April 16, 2018, 10:43:03 AM by Gregg Bell »

"When people agree with me I always feel that I must be wrong." Oscar Wilde
Gregg Bell | Website | Amazon author page | Blog | Twitter | Google+ | Facebook

Offline BigSlimJim

  • Status: Dr. Seuss
  • *
  • Posts: 17
    • View Profile
Re: A New AMS Thread
« Reply #721 on: April 16, 2018, 12:43:11 PM »
2.  I believe ad copy is where 'relevance' comes in.  It is certainly judged for relevance when being approved, but I would think its relevance would continue to be evaluated as time progresses.  But evaluated per associated keyword, not evaluated per ad overall.
I spoke with a KDP rep today who said that the auction algorithm does not consider the ad copy. If the ad copy is found to be compliant with the Book Ads Creative-acceptance Policy (a process that I think is about advertising rules, not relevance), and the ad is approved, then the individual keyword impressions are determined by keyword relevance to searches, bid price, CTR history (if there's a history), etc. Naturally, the ad copy always pays a huge part in every keyword's click-through rates. If shoppers aren't intrigued by an ad, they don't click, and the CTR falls. But the AMS algorithm doesn't directly look at the ad copy to match a search and provide a sponsored impression, according to this rep and my earlier suspicions.
However, somebody, somewhere, does know exactly how the AMS algorithms work.
But, so far, they're not talking.
You're more confident than I am that such a somebody exists :-). The more I think I've learned about how the reactive (not thinking) robot operates, the harder it is to comprehensively retain and explain. And I know next to nothing, compared to what that somebody would have learned. There are probably many more than three or four factors that the robot is looking at to make the most profitable choice (For Amazon) between competing bidders. Other factors might include sales rank, trending, and sale price, just off the top of my head. To make us dreamers fully aware of all the nuances involved in the algo's choices might be like a casino teaching blackjack players how to count cards. Not a wise investment of the house's energies.

Offline BigSlimJim

  • Status: Dr. Seuss
  • *
  • Posts: 17
    • View Profile
Re: A New AMS Thread
« Reply #722 on: April 16, 2018, 01:00:25 PM »
Cassie, I have a couple questions regarding how you pause keywords that don't fulfill your desired minimums of 1 to 2 clicks per thousand impressions and one or more sales per every ten clicks:

1. Do you pause a click-less keyword as soon as it reaches a thousand impressions, or do you give the word a few thousands to show an average? If the latter, at how many thousands do you usually judge?

2. Similarly, do you pause a no-sale keyword at the 10-click event? If otherwise, how many clicks do you allow to determine a more-averaged sales-to-click ratio?

Bonus question to you or others: Can you give a no-click keyword all the time it needs to get a thousand impressions? How long is too long with slow-to-show words that might have promise over time?

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #723 on: April 16, 2018, 01:01:00 PM »
So it seems to me, if you're content to be on the back pages you can have a maintain a modest bid, but if you increase your bid enough to move up to the front page, you're likely to get hit with a big cost for a click. (Even with smallish authors.) Would this be accurate?

And yet, sometimes, even with a high bid, the ACPC stays the same (with more clicks).

It seems so to me too Greg. Depends on what the algo bot sees that it can do to place the ad - bid of 75 cents can turn up costing only 14 cents too CPC but likely 75 cents as you are bidding to be first on the carousel it seems to me at 75 cent bid. Depends on other bidders and the key word.

Up front on the carousel is going to cost you. You are likely to get more clicks too I suspect. You have the first, "Wow! Let's see what this is about."  And they move on likely leaving you with the bill. 

So for a new author like me and not so many other works, you are a hobbyist. This is going to cost you to get read. Really it is not a lot of money. But KU book coffers are full I suspect and they are super picky it seems.

So if your work is not Hemmingway or King - stay with 15 cent top bid and appear on the back of the carousel and your costs are just hobby costs it seems to me. You may never hit a $6 a day budget at 15 cent bid.  I bet even at 75 cents you won't either - depends on the number of keywords and if they all have that high bid. But you may feel a lot more pain $ wise if you have a high bid and budget and lots of keywords that get impressions. Readers can click you to death and say, "Nah," as the look it over.

If you write a good story, it seems you need to give it a lot of time, a lot more than I imagined when I tried this 8 years ago. The mass of books is the problem for those just now getting into the game (hobby). Unreal now it seems to me.

So give it time and a fair bid and budget I can live with. Flesh out the book to 250 pages from 78 and get some AMS wiggle room. But after awhile, even with a fair number of clicks, and a few buys or borrows, I drop this hobby. It was fun, but not worth the money and time I am putting into it. I have as much fun reading and leaving reviews.  I pick up plenty of free books, more than I can read, and I am not even KU. 

If you like reading and commenting on the book, Amazon has a ton of free stuff.
« Last Edit: April 16, 2018, 01:30:41 PM by Max 007 »

Offline Max 007

  • Status: Jane Austen
  • ***
  • Posts: 257
  • Gender: Male
    • View Profile
Re: A New AMS Thread
« Reply #724 on: April 16, 2018, 01:23:10 PM »
I spoke with a KDP rep today who said that the auction algorithm does not consider the ad copy. If the ad copy is found to be compliant with the Book Ads Creative-acceptance Policy (a process that I think is about advertising rules, not relevance), and the ad is approved, then the individual keyword impressions are determined by keyword relevance to searches, bid price, CTR history (if there's a history), etc. Naturally, the ad copy always pays a huge part in every keyword's click-through rates. If shoppers aren't intrigued by an ad, they don't click, and the CTR falls. But the AMS algorithm doesn't directly look at the ad copy to match a search and provide a sponsored impression, according to this rep and my earlier suspicions.You're more confident than I am that such a somebody exists :-). The more I think I've learned about how the reactive (not thinking) robot operates, the harder it is to comprehensively retain and explain. And I know next to nothing, compared to what that somebody would have learned. There are probably many more than three or four factors that the robot is looking at to make the most profitable choice (For Amazon) between competing bidders. Other factors might include sales rank, trending, and sale price, just off the top of my head. To make us dreamers fully aware of all the nuances involved in the algo's choices might be like a casino teaching blackjack players how to count cards. Not a wise investment of the house's energies.

I don't think this is like gambling.  But as you point out, let's say push-come- to-shove you have a bid equal to another ad, algo bot I bet does look at the ASIN rank, history, reviews, etc (info that is already available to the bot and in the system) and bumps you or the other deeper into the carousel.

The best answer you will get from Zon is it is a proprietary algo. But we know it does and may have to do some considering beyond what we initially think on the surface. I just don't think it keeps a history on the ads performance WRT keywords (that is a lot of frivolous and erratic data to me). But it may rank the bid differently than what we hope against other close bids to ours and bump the other or ours further down the carousel so to speak.

There is a lot of info the bot does have access to, to make a determination, and you brought it up. And it will always likely default to what is likely to make Zon money --- if there is a choice of choices that appear equal at first glance to the algo bot.

If it can place your ad, I think it always will.  That is the AMS promise sort of, in a way, ... to you, ... maybe ...  :)

I've turned my ad off for now. I will flesh out the book to 250 pages for more wiggle room to mess with AMS better.

I will know soon enough if I am spinning my wheels. 
« Last Edit: April 16, 2018, 01:25:35 PM by Max 007 »